SEARCH

SEARCH BY CITATION

Keywords:

  • Addiction;
  • cocaine;
  • gender;
  • genetics;
  • opioids;
  • PDYN

Abstract

  1. Top of page
  2. Abstract
  3. Materials and methods
  4. Results
  5. Discussion
  6. References
  7. Acknowledgments

Genetic factors are believed to account for 30–50% of the risk for cocaine and heroin addiction. Dynorphin peptides, derived from the prodynorphin (PDYN) precursor, bind to opioid receptors, preferentially the kappa-opioid receptor, and may mediate the aversive effects of drugs of abuse. Dynorphin peptides produce place aversion in animals and produce dysphoria in humans. Cocaine and heroin have both been shown to increase expression of PDYN in brain regions relevant for drug reward and use. Polymorphisms in PDYN are therefore hypothesized to increase risk for addiction to drugs of abuse. In this study, 3 polymorphisms in PDYN (rs1022563, rs910080 and rs1997794) were genotyped in opioid-addicted [248 African Americans (AAs) and 1040 European Americans (EAs)], cocaine-addicted (1248 AAs and 336 EAs) and control individuals (674 AAs and 656 EAs). Sex-specific analyses were also performed as a previous study identified PDYN polymorphisms to be more significantly associated with female opioid addicts. We found rs1022563 to be significantly associated with opioid addiction in EAs [P = 0.03, odds ratio (OR) = 1.31; false discovery rate (FDR) corrected q-value]; however, when we performed female-specific association analyses, the OR increased from 1.31 to 1.51. Increased ORs were observed for rs910080 and rs199774 in female opioid addicts also in EAs. No statistically significant associations were observed with cocaine or opioid addiction in AAs. These data show that polymorphisms in PDYN are associated with opioid addiction in EAs and provide further evidence that these risk variants may be more relevant in females.

The United States National Survey on drug use and health showed that in 2009, there were 1.1 million people aged 12 or older who were classified as dependent on or abusing cocaine and 399,000 individuals classified as dependent on or abusing heroin (2009 National Survey on Drug Use and Health 2010), http://www.oas.samhsa.gov. The genetic liability for cocaine and/or heroin addiction is estimated to be 30–50% (Kendler et al. 2003; Tsuang et al. 1998); however, finding susceptibility genes has been difficult due to the complex mode of inheritance and multiple genes that contribute to increased disease risk. Genes encoding proteins that are involved in opioidergic neurotransmission are highly relevant for both heroin and cocaine dependence as the endogenous opioid system mediates the effects of exogenously administered opiates and cocaine (Kreek et al. 2005; Steiner & Gerfen 1993).

Dynorphins, opioid peptides derived from the prodynorphin (PDYN) precursor (Schwarzer 2009), bind to all three opiate receptors, but exhibit considerable preference for the kappa-opioid receptor (KOR). Dynorphins are believed to mediate the aversive effects of drugs of abuse as experimental administration of KOR agonists in animals produces place aversion (Mucha & Herz 1986; Shippenberg & Herz 1986; Zhang et al. 2005) and dysphoria/confusion in humans (Shippenberg et al. 2007; Walsh et al. 2001). This is believed to be due, in part, to a reduction in dopaminergic neurotransmission on KOR stimulation and increased PDYN gene expression (Di Chiara & Imperato 1988; Nestler 2004). Exposure to cocaine upregulates dynorphin immunoreactivity in brain regions such as the caudate and ventral palladium (Frankel et al. 2008; Hurd & Herkenham 1995; Staley et al. 1997) and chronic exposure to heroin increases PDYN mRNA in the central amygdala and nucleus accumbens shell (Solecki et al. 2009). Naloxone-precipitated withdrawal was found to accentuate the increase in morphine-induced dynorphin expression in the striatum and accumbens in rats (Nylander et al. 1995). Such studies have led researchers to hypothesize that dynorphin may contribute to the negative emotional states experienced during withdrawal from drugs of abuse and motivate continued drug use as a consequence of negative reinforcement (Koob & Le Moal 2008; Wee & Koob 2010).

Numerous studies have analyzed the association between polymorphisms in PDYN and both cocaine and opioid addiction, with varying results. Several studies report positive associations with opioid and cocaine addiction (Clarke et al. 2009; Dahl et al. 2005; Flory et al. 2011; Yuferov et al. 2009); however, negative findings have also been reported (Chen et al. 2002; Nikoshkov et al. 2008; Ray et al. 2005; Williams et al. 2007). Many of the variants in PDYN associated with increased risk for drug addiction in PDYN have also been shown to have functional consequences for gene expression. For example, individuals harboring three or four copies of a 68-bp variable nucleotide tandem repeat (VNTR) in the PDYN promoter exhibit greater PDYN expression than those with only one or two copies (Nikoshkov et al. 2008; Zimprich et al. 2000). Importantly, this result has not been consistently replicated and subsequent studies have found expression driven by the 68-bp repeat to be dependent on cell type (Rouault et al. 2011). Single nucleotide polymorphisms (SNPs) in the 3′untranslated region (UTR) are also associated with reduced PDYN expression, as determined by allelic imbalance studies in postmortem brain samples (Yuferov et al. 2009). Furthermore, a comprehensive analysis of SNPs in the promoter of PDYN, using both in vitro and in vivo techniques across different brain regions, found several different variants influencing PDYN expression. SNP rs1997794 was found to explain most of the variance in gene expression using both techniques (Babbitt et al. 2010).

The purpose of this study was to examine the relevance of potentially functional PDYN genetic variation in cocaine and opioid addiction in European Americans (EAs) and African Americans (AAs). rs1997794, rs1022563 and rs910080 were selected for genotyping. rs1997794, located in the promoter, and rs910080 in the 3′UTR were selected as they have previously been shown to be associated with cocaine and opioid dependence and have been shown to be associated with differential PDYN gene expression (Babbitt et al. 2010; Clarke et al. 2009; Yuferov et al. 2009). rs1022563 was also selected for genotyping as this SNP was found to be significantly associated in a study on female Chinese heroin addicts (Clarke et al. 2009). Furthermore, as a previous study has shown that the association of PDYN SNPs with opioid addiction may be more relevant in females (Clarke et al. 2009), a secondary analysis was performed with a focus on female cocaine/opioid addicts.

Materials and methods

  1. Top of page
  2. Abstract
  3. Materials and methods
  4. Results
  5. Discussion
  6. References
  7. Acknowledgments

SNP genotyping

rs910080, rs199794 and rs1022563 were genotyped in PDYN and in cocaine- and opioid-addicted subjects and controls. SNP genotyping was performed using Taqman® SNP Genotyping Assays (Applied Biosystems Inc., Foster City, CA, USA) as per the manufacturer's protocol. Quality control was maintained by genotyping 10% duplicates, which were checked for genotype concordance across the populations. The duplicate concordance rate for all three SNPs was 100%.

Subject information

Cocaine- and opioid-addicted individuals

DNA samples were requested and acquired through the National Institute of Drug Addiction (NIDA) Center for Genetic Studies in conjunction with the Washington University and the Rutgers University Cell & DNA Repository. Opioid-dependent samples were acquired from the NIDA Repository Studies 1 (PI: J. Gelernter et al.) and 5 (PI: M.J. Kreek), and cocaine-dependent samples were acquired from Studies 7 (PI: L. Bierut) and 13 (PI: J. Cubells). Opioid-addicted (EA: n = 1041; male 66.1%; AA: n = 284 male 68%) and cocaine-addicted subjects (EA: n = 336; male 50.3%; AA: n = 1248; male 62%) of EA and AA decent met Diagnostic and Statistical Manual of Mental Disorders-IV (DSM-IV) criteria for dependence.

A portion of the AA cocaine-addicted subjects (n = 908) were collected during clinical studies for cocaine addiction treatment at the University of Pennsylvania Treatment Research Center. Subjects were at least 18 years of old. All were assessed with the Structured Clinical Interview for DSM Disorders and urine drug screens were obtained. All patients had a clinical diagnosis of cocaine addiction as defined by DSM-IV. Family history was not obtained and ethnicity was determined by self-report. All psychiatric axis-I disorders except alcohol dependence/abuse and nicotine dependence were used as exclusion criteria. In addition, participants were excluded if they had a history of a seizure disorder (except cocaine-induced seizures) or a severe medical illness, including a history of acquired immune-deficiency syndrome (but not merely of human immunodeficiency virus+ status). Individuals currently being treated with psychotropic medications or with psychiatric symptoms, including psychosis, dementia, suicidal or homicidal ideation, mania or depression requiring antidepressant therapy were also excluded. For all samples, genomic DNA was extracted from peripheral leucocytes within obtained blood samples by standard protocols. All protocols were approved by the Institutional Review Boards at the Thomas Jefferson University and the University of Pennsylvania, and all subjects provided written informed consent before blood sample collection.

Control individuals

EA control individuals (n = 656; male = 50.8%) and AA control individuals (n = 674; male = 37.3%) were acquired from the National Institute of Mental Health Genetics Initiative (NIMH-GI; www.nimhgenetics.org). Control subjects were screened using an online self-report clinical assessment, which screens for adult psychiatric diseases using a modified version of the Composite International Diagnostic Interview–Short Form. Individuals who self-identified as having an axis-I psychiatric disorder were excluded from this study. A portion of AA controls (n = 92) were collected simultaneously with the cocaine-dependent patients at the University of Pennsylvania and were also genotyped. Cases and controls used in this study are summarized in Table 1.

Table 1.  Number of individuals for cocaine and opioid addicts and controls
PopulationCocaine-addicted casesOpioid-addicted casesControls
  1. Table showing number of individuals in each population, % of male individuals for each population displayed in parentheses and mean age and standard deviation (1 SD) displayed in [ ].

European Americans336 (50.3%) [36.13 ± 8.5]1046 (66.1%) [37.89 ± 11.8]656 (50.8%) [52.98 ± 17.6]
African Americans1248 (62%) [41.97 ± 6.8]284 (68%) [49.7 ± 8.86]674 (37.3%) [45.8 ± 14.6]
Statistical analysis

The allelic and genotypic association of SNPs with opioid and cocaine addiction were determined using the chi-squared test and model tests in the software package PLINK v1.04 (Purcell et al. 2007). For the model tests, genotypic (2df; AA vs. AB vs. BB), recessive (1df; AA,AB vs. BB) and dominant (1df; AA vs. AB,BB) tests were performed. However, the dominant and recessive tests are only reported if the genotypic tests displayed trend toward significance (P < 0.06). Owing to the different minor allele frequencies of the polymorphisms in EA and AA populations, the associations of the two ethnicities were analyzed separately. To conduct female-specific analyses, male cases and controls were removed and female cases compared with female controls for each SNP.

The program QVALUE 1.0 was used (Storey 2002; Storey et al. 2004; Storey & Tibshirani 2003) to correct for multiple testing on all reported P-values. This includes allelic and genotypic P-values for the total population (males and females) and females separately across both ethnicities (experiment-wise correction). A q-value <0.05 was considered statistically significant.

Linkage disequilibrium (LD) analysis was performed for the AA and EA populations separately, but combining cases and controls, using the program HAPLOVIEW (Barrett et al. 2005). Haplotype blocks are defined according to the method defined by Gabriel et al. (2002). LD analysis was also performed for cases and controls separately; however, the LD was not found to be substantially different between the two populations (data not shown). Sliding window haplotype analysis was performed in PLINK using the expectation maximization algorithm implicated in the software packages (http://pngu.mgh.harvard.edu/purcell/plink/; Purcell et al. 2007). Permutation analysis of the haplotypic associations performed using PLINK and 10 000 permutations per haplotype were conducted. Haplotype analysis did not add any further significance to the findings (data not shown).

A power analysis was conducted using Genetic Power Calculator (Purcell et al. 2003). Using a dominant test, this study had 81% power to detect an association with an odds ratio (OR) of 1.3 and a minor allele frequency (MAF) of 30% (α = 0.05). Using recessive, genotypic and allelic tests, the power to detect association was 12%, 72% and 71%, respectively (α = 0.05).

Results

  1. Top of page
  2. Abstract
  3. Materials and methods
  4. Results
  5. Discussion
  6. References
  7. Acknowledgments

The MAF for each of our SNPs in the control population was comparable to the MAFs reported in other studies (Table 2). The frequency of the minor allele in AA and EA populations is higher in our sample than in that reported by Yuferov et al. (2009) (3–7%); however, we have a much larger sample size as they only genotyped 76 AA controls and 65 EA controls. For the EA population, our MAFs are more similar to that of Xuei et al. (2006) as only a 2% difference is observed. There are no studies published, to our knowledge, which have examined rs1022563 in an EA or AA population.

Table 2.  Comparison of SNP minor allele frequency (MAF) between control populations across different studies
SNP (minor allele)Control population Yuferov et al. (2009) Xuei et al. (2006) HapMap
  1. MAF for each SNP in the control population compared with the MAF in control populations recorded by Yuferov et al. (2009) and Xuei et al. (2006). MAF are shown (www.hapmap.org) for the CEU (Utah residents with Northern and Western European ancestry from the CEPH collection) and ASW (African ancestry in Southwest USA) populations.*The MAF for rs1997794 is T in African American populations and C in European populations.

African Americans
 rs1022563 (T)0.19
 rs910080 (G)0.480.430.5
 rs1997794* (T)0.250.220.29
European Americans
 rs1022563 (T)0.170.11
 rs910080 (G)0.280.210.260.26
 rs1997794* (C)0.380.320.360.38

LD analysis was performed between the three SNPs in EA and AA populations (Fig. 1a and b). A haplotype block between rs1022563 and rs910080 was observed in both populations (D′ > 0.98), suggesting that these polymorphisms occurred on the same chromosome and little recombination has subsequently occurred in individuals of either European or African ancestry. There is moderate LD observed between rs1997794 and rs910080 in the EA population (D′ = 0.9); however, the LD in the AA population is low (D′ = 0.28), suggesting that recombination has occurred between these two SNPs in AAs.

image

Figure 1. Linkage disequilibrium (LD) between genotyped PDYN SNPs. (a) LD between three SNPs genotyped in EAs (using genotype data from current study, cocaine and opioid-addicted individuals and controls collated). (b) LD between three SNPs genotyped in African Americans (using genotype data from current study, cocaine- and opioid-addicted individuals and controls collated). The shading represents r2 and numbers inside the boxes refer to D′. Haplotype blocks defined according to the method of Gabriel et al. (2002). LD data are visualized using HAPLOVIEW (Barrett et al. 2005).

Download figure to PowerPoint

Chi-squared allelic association analysis showed a significant association of rs1022563 in EA opioid addicts compared with controls [χ2 = 7.41, 1df, P = 0.006, OR (95% CI) 1.31 (1.08–1.6)], which was also observed at the genotypic level (χ2 = 8.3, 1df, P = 0.004 – dominant test for association), both these tests withstood correction for multiple testing (q-value <0.05). No significant associations were observed for rs910080 or rs1997794 at the allelic or genotypic level in EA opioid addicts (P > 0.08; Table 3). Chi-squared association analysis also showed no significant association of any of the PDYN SNPs in AA opioid addicts compared with controls (P > 0.17; Table 3). An analysis of PDYN polymorphisms and cocaine addiction in EA and AA populations also showed no significant associations (Table 4) at the allelic or genotypic level (P > 0.11). Each SNP was tested for deviation from Hardy–Weinberg equilibrium (HWE) in controls. No significant deviations from HWE were found (data not shown).

Table 3.  Genetic association analysis of PDYN SNPs and opioid addiction
SNPPopulation111222EA allelic P-valueEA genotypic P-valueOR (95% CI)111222AA allelic P-valueAA genotypic P-valueOR (95% CI)
  1. Allelic and genotypic chi-square associations with PDYN SNPs and opioid addiction in European Americans (EAs) and African Americans (AAs). 11, minor homozygotes; 12, heterozygotes; 22, major homozygotes, genotype N are presented with genotype frequencies presented in parentheses; OR, odds ratios; 95% CI, confidence intervals. Bold P-values remain significant after testing for multiple correction (q-value < 0.05; (Storey 2002; Storey et al. 2004; Storey & Tibshirani 2003).

  2. P-value for dominant test for association.

rs1022563Cases14 (1.4%)228 (24.1%)706 (74.5%) 0.006 0.02 (0.004)1.31 (1.08–1.6)17 (5%)111 (33%)208 (62%)0.170.211.18 (0.92–1.51)
 Controls12 (1.9%)195 (30.3%)437 (67.8%)   20 (3%)210 (31.8%)430 (65.2%)   
rs910080Cases62 (6%)391 (38%)577 (56%)0.090.091.15 (0.98–1.34)76 (23%)157 (47.6%)97 (29.4%)0.560.551 (0.82–1.22)
 Controls57 (8.8%)242 (37.6%)345 (53.6%)   151 (23%)337 (51%)175 (26%)   
rs1997794Cases121 (11.7%)483 (46.6%)433 (41.8%)0.080.091.13 (0.98–1.31)29 (8.6%)132 (39.3%)175 (52.1%)0.350.621.11 (0.89–1.39)
 Controls100 (15.3%)294 (45.1%)258 (39.6%)   42 (6.3%)251 (37.6%)375 (56.1%)   
Table 4.  Genetic association analysis of PDYN SNPs and cocaine addiction
SNPPopulation111222EA allelic P-value (1df)EA genotypic P-value (2df)OR (95% CI)111222AA allelic P-value (1df)AA genotypic P-value (2df)OR (95% CI)
  1. Allelic and genotypic chi-square associations with PDYN SNPs and cocaine addiction in European Americans (EAs) and African Americans (AAs). 11, minor homozygotes; 12, heterozygotes; 22, major homozygotes, genotype N are presented with genotype frequencies presented in parentheses; OR, odds ratios; 95% CI, confidence intervals.

rs1022563Cases9 (2.7%)78 (23.8%)241 (73.4%)0.180.081.2 (0.92–1.55)48 (4.2%)320 (27.9%)777 (67.9%)0.490.110.94 (0.79–1.12)
 Controls12 (1.9%)195 (30.3%)437 (67.8%)   20 (3%)210 (31.8%)430 (65.2%)   
rs910080Cases16 (6.6%)90 (36.9%)138 (56.5%)0.260.491.15 (0.9–1.45)287 (25.4%)531 (47.1%)310 (27.5%)0.710.31.03 (0.9–1.18)
 Controls57 (8.9%)242 (37.6%)345 (53.6%)   151 (23%)337 (51%)175 (26%)   
rs1997794Cases40 (12.4%)152 (47.1%)131 (40.5%)0.40.461.09 (0.89–1.32)78 (6.7%)434 (37.3%)650 (56%)0.820.951.02 (0.87–1.19)
 Controls100 (15.3%)294 (45.1%)258 (39.6%)   42 (6.3%)251 (37.6%)375 (56.1%)   

To test the hypothesis that PDYN polymorphisms increase the risk for addiction more strongly in females, female cases and controls were compared separately using chi-squared tests for allelic and genotypic association. In female EA opioid addicts (355 cases vs. 323 controls), significant associations were observed for all three PDYN SNPs tested. rs1022563 was associated at the allelic level (χ2 = 6.91, 1df, P = 0.009) and the genotypic level (2df, P = 0.01) and withstood correction for multiple testing. This association is weaker statistically than when males and females were analyzed together; however, this is a reflection of reduced sample size rather than a diminished effect size, as the OR is increased for rs1022563 [OR (95% CI) = 1.51 (1.11–2.05)]. rs910080 is significantly associated in female EA opioid addicts at the allelic level; however, only nominal significance is observed (χ2 = 4.03, 1df, P = 0.04, OR (95% CI) = 1.28 (1.01–1.64)). rs1997794 shows a trend toward significance at the genotypic level in female EA opioid addicts (2df, P = 0.06) and is nominally significant when recessive tests for association are performed (1df, P = 0.02). The OR for rs1997794 is also increased when female opioid addicts are analyzed separately [OR (95% CI) = 1.33 (1.06–1.68)]. Detailed allelic and genotypic associations of EA female opioid addicts vs. controls are summarized in Table 5. To determine whether females were responsible for the association of rs1022563 and EA opioid addiction in the total cohort (males and females), we performed exploratory association analysis in males. We did not find males to show an association between rs1022563 and addiction and can therefore conclude that it is the female part of the sample driving this association (data not shown). No statistically significant associations were observed when female AA opioid addicts or EA/AA cocaine addicts were analyzed separately (data not shown).

Table 5.  Genetic association analyses of PDYN SNPs and female opioid addiction in EAs
SNPPopulation111222Allelic P-value (1df)Genotypic P-value (2df)OR (95% CI)
  1. Showing allelic and detailed genotypic associations for opioid dependence in EA females (355 cases vs. 323 controls). 11, minor homozygotes; 12, heterozygotes; 22, major homozygotes. Bold P-values remain significant after correction for multiple testing (q-value < 0.05; Storey 2002; Storey et al. 2004; Storey & Tibshirani 2003).

  2. Recessive test for association.

rs1022563Cases1 (0.31%)80 (24.84%)241 (74.84%) 0.009 0.01 1.51
 Controls6 (1.9%)102 (32.3%)208 (65.8%)  (1.11–2.05)
rs910080Cases17 (4.9%)133 (38.2%)198 (56.9%)0.040.071.28
 Controls29 (9.1%)126 (39.5%)164 (51.4%)  (1.01–1.64)
rs1997794Cases37 (10.5%)165 (46.9%)150 (42.6%)0.210.061.33
 Controls53 (16.5%)133 (41.4%)135 (42.1%) (0.02)(1.06–1.68)

Each SNP was tested for deviation from HWE in cases and controls in the female portion of the cohort. No significant deviations from HWE were observed except for rs1022563 in EA opioid-addicted females where the P-value was 0.03. However, as a significant association was found in this population, it may be the association with addiction, which is causing this deviation. A significant deviation was also found in EA controls for rs1997794 as the HWE P-value was 0.04.

Discussion

  1. Top of page
  2. Abstract
  3. Materials and methods
  4. Results
  5. Discussion
  6. References
  7. Acknowledgments

The results from this study provide further evidence for an association between PDYN polymorphisms and opioid addiction in females. Thus, while rs1022563 was found to be associated with opioid addiction in EAs overall (genotypic P = 0.004, OR = 1.3), the OR increased to 1.5 when females were analyzed separately, suggesting a greater influence of this SNP in females. Similarly, rs910080 and rs1997794 were not found to be significantly associated in the total EA population; however, when females were analyzed separately, rs910080 was found to be nominally significant at the allelic level (P = 0.04) and rs1997794 at the genotypic level (P = 0.02, recessive test for association). The OR increased for both these SNPs (from 1.15 and 1.13 to 1.28 and 1.32, respectively), again showing a greater effect size in females. However, a deviation from HWE was observed in female EA controls for rs1997794. Caution should therefore be taken when interpreting these data as deviation from HWE in controls may indicate genotyping error; however, it should be noted that EA controls were in HWE in the total population (P = 0.28), and therefore, the deviation from HWE may be an artifact of splitting and retesting the sample. We did not find evidence of positive association between these three SNPs and either opioid addiction in an AA population or cocaine addiction in the EA and AA population.

These data support the earlier findings of Clarke and colleagues who found rs1997794 and rs1022563 to be associated with opioid dependence in a Chinese population, specifically in females (Clarke et al. 2009). The C and T alleles of rs1022563 and rs1997794, respectively, are associated with risk in female opioid addicts in the Chinese population, which is confirmed by our present data, despite the large differences in MAF between the two populations. The minor allele is switched for these two SNPs in Chinese and European populations; however, we find the same association between PDYN risk alleles and opioid addiction, further supporting the hypothesis that these variants are highly relevant for influencing opioid use in females.

Other studies have analyzed PDYN in substance-addicted populations, although to our knowledge, these studies did not perform gender-specific analysis, with the exception of Clarke et al. (2009). Wei et al. (2011) found the 68-bp VNTR and a three-SNP haplotype containing rs1022563 to be associated with opioid addiction in a Chinese population. Furthermore, rs1997794 and rs910080 were found to be associated with alcoholism using a large EA sample (Xuei et al. 2006). Other studies on PDYN polymorphisms and substance addiction, however, were not replicated by our data. A study on cocaine and cocaine/alcohol codependent EA and AA populations found rs910080 to be significantly associated with cocaine addiction in EAs (Yuferov et al. 2009). We were not able to replicate this finding as rs910080 was not found to be significantly associated with cocaine addiction in our EA population (P≥ 0.26). Studies examining the putative functional 68-bp repeat have found positive associations with opioid addiction. A weak association of the VNTR was observed in an AA opioid-addicted population (Ray et al. 2005) and an association with CD in AAs has also been shown (Dahl et al. 2005). We were not able to find an association of PDYN polymorphisms in either our opioid or cocaine AA populations; however, this may be because the risk conferred by the 68-bp repeat is distinct from that conferred by the SNPs analyzed in this study.

Potential biological function has been ascribed to two of the SNPs analyzed in this study; rs1997794 and rs910080. The promoter SNP, rs1997794, was found to be associated with differential gene expression in the prefrontal cortex, occipital cortex and temporal cortex with the T allele associated with lower expression of PDYN (Babbitt et al. 2010). rs910080, located in the 3′UTR, was found to be part of a haplotype associated with altered PDYN expression, and the G allele of this SNP was associated with decreased expression in the dorsal and ventral striatum. The T allele of rs1997794 was found to be the risk allele for opioid addiction in our study; however, the G allele of rs910080 was found to confer protection against opioid addiction. Therefore, it is unclear whether lower PDYN gene expression is a risk factor for opioid addiction; however, it is possible that SNPs influence gene expression differently across different brain regions. Indeed, it is likely that many cis-acting variants influence expression of PDYN, and the transcriptional control of PDYN has already been shown to vary according to genotype, cell type and sex (Babbitt et al. 2010; Rouault et al. 2011).

One limitation of this study is that ancestry informative markers were not genotyped in the EA or AA populations, and therefore, we cannot rule out population stratification as the cause of the associations observed. However, as we were able to replicate the findings of Clarke et al. (2009), in a different ethnic group (with opposite minor alleles at rs1997794 and rs1022563), this suggests that these alleles at least may increase risk for opioid addiction in females. Another limitation pertains to the use of DNA samples stored at a central repository. The numbers of cocaine- and opioid-addicted persons used in this study differs greatly between EA and AA populations. This is a technical artifact, as the maximum number of cases was used for both ancestral populations; however, the numbers reflect the resources available at the time and not the prevalence of a particular form of addiction in the two populations. The majority of DNA samples from drug-addicted individuals used in this study are from the NIDA repository, which stores DNA samples for genetic studies on addiction and are available to researchers on request. Under the current system, it is not possible to ascertain which exact DNA samples are used for each study, and therefore, caution should be exercised when attempting to replicate findings from groups that have either used NIDA DNA samples or have contributed DNA samples to the NIDA repository. To our knowledge, the only studies that used these samples to examine PDYN polymorphisms are by Levran et al. (2008, 2009), who examined EA and AA populations respectively. They used opioid-addicted DNA samples, which were subsequently submitted to the NIDA repository, to analyze 1350 variants across a number of candidate genes, including PDYN (Levran et al. 2008, 2009). rs1997794 was analyzed in these studies and found not to be associated with opioid addiction; our study cannot therefore be considered a true replication of this finding as there is likely to be some overlap of DNA samples between the two studies. The issue of using repository samples should also be considered when attempting meta-analyses as individuals should not be duplicated within such analyses as overrepresentation of ‘addiction’ alleles may occur to generate false-positive results.

When selecting control subjects for any genetics study, it is important to control for as many environmental factors as possible. Therefore, when selecting controls for drug addiction studies, it may be beneficial to use control subjects who have had exposure to the drug in question as, in the case of drug addiction, the most important environmental factor is exposure to the drug itself (Buckland 2001). This can be difficult when studying addiction to illicit drugs as a substantial portion of the population will not have had any exposure to the drug being studied. Control individuals may therefore carry risk alleles for drug addiction; however, they are not addicted, having never been exposed to the drug in question. The effect size of the variant may therefore be underestimated or undetected altogether. However, the addicted case population will be enriched for addiction alleles compared with controls, and therefore, it is still worthwhile to make the comparison between the two. Furthermore, the risk alleles in question may increase propensity for drug addiction in general by promoting sensitivity to drug reward or sensation-seeking behaviors.

The control subjects used in this study were obtained from the NIMH-GI repository, which stores DNA samples from individuals who had been screened for psychiatric disorders using an online questionnaire. There are limitations to using online questionnaires to screen for psychiatric diagnoses, as in the NIMH-GI control sample, the rate of psychiatric diagnoses was found to be higher than in the actual population (Sanders et al. 2010). This may be because individuals place more emphasis on experiences that a psychiatrist would rule out as pathological. Other limitations include the sampling bias of individuals with an email address and concerns regarding the integrity of answers given. Different studies may have used different exclusion/inclusion criteria for their control subjects. The controls used by Yuferov et al. (2009) were screened for psychiatric disorders and also had much stricter exclusion criteria regarding drug use, as those individuals who had drunk to intoxication in the past month or used illicit drugs in the past month were excluded. Such differences should be considered when studies fail to replicate each other, as the groups in question may be quite different in terms of exposure to drugs of abuse and presence/absence of psychiatric disorders.

The false discovery rate (FDR) method employed in QVALUE (Storey 2002; Storey et al. 2004; Storey & Tibshirani 2003) was used to correct for multiple testing in this study and as such four P-values remained significant. Had we used the more stringent Bonferroni correction, no SNPs would have remained significant. Forty-eight tests were carried out in total, leading to a threshold for significance set at 0.001 (0.05/48); our most significant P-value was 0.004. However, most interestingly in this study was the finding that the OR for each of the risk alleles in EA opioid addiction was increased in females. Owing to the smaller sample size of the female cohort the P-values are larger; however, the finding points toward an area for further research when considering the role of PDYN in opioid addiction.

One question raised by our findings is why would polymorphisms in PDYN be more relevant for opioid addiction in females? The primary target for exogenous opioids in the brain is the mu-opioid receptor (MOR). KOR (the primary receptor for PDYN) has been found to form MOR/KOR heterodimers in the spinal cord of rats (Chakrabarti et al. 2010). These heterodimers show higher expression in proestrus female rats compared with male rats and this leads to sexual dimorphism in the antinociceptive effects of morphine (Liu et al. 2011). Similar findings are observed in humans as studies have shown women report greater analgesia from KOR/MOR ligands such as pentazocine, nalbuphine and butorphanol (Gear et al. 1996a,b). Sex differences have also been observed at the level of PDYN gene expression in humans. A study on PDYN promoter polymorphisms in different brain regions and cell types found significantly different levels of gene expression dependent on both genotype and sex (Babbitt et al. 2010).

Few studies with a focus on females have been conducted at the intersection of the dynorphin–KOR system and drugs of abuse. However, a study on Pdyn null mice found that reduced alcohol consumption and place preference was observed, but only in null females (Blednov et al. 2006). Together these data suggest that there are sex differences in the dynorphin–KOR system, which may be relevant for sexually dimorphic responses to morphine and alcohol, and that these responses may also be influenced by genotype in a sex-dependent manner.

As exposure to opioids and cocaine has been shown to induce neuroadaptive upregulation of PDYN expression (Frankel et al. 2008; Hurd & Herkenham 1995; Nylander et al. 1995; Solecki et al. 2009; Staley et al. 1997), it has been hypothesized that KOR antagonists could be effective drug therapies for addiction (Shippenberg et al. 2007). Indeed, trials of buprenorphine (partial MOR agonist/KOR antagonist) in combination with naltrexone have shown better treatment outcomes for opioid dependence compared with treatment with naltrexone alone (Gerra et al. 2006). Future work would be to determine whether these treatment regimens are more effective in females and also whether PDYN genotype helps predict treatment response to such pharmacotherapies. Furthermore, additional genotyping of independent opioid-addicted cohorts should be carried out, focusing on females only, to confirm the role that these polymorphisms may play in female-specific opioid addiction.

References

  1. Top of page
  2. Abstract
  3. Materials and methods
  4. Results
  5. Discussion
  6. References
  7. Acknowledgments

Acknowledgments

  1. Top of page
  2. Abstract
  3. Materials and methods
  4. Results
  5. Discussion
  6. References
  7. Acknowledgments

This research was supported by NIMH Grant K08MH080372 (FWL) and the Center for Neurobiology and Behavior, Department of Psychiatry, University of Pennsylvania, Training Program in Neuropsychopharmacology (T32MH014654, P.I.: I. Lucki) and a Distinguished International Scientist Award. Financial support is gratefully acknowledged from NIDA Grant P20DA025995 (P.I.: W. Berrettini), the Veterans Administration Mental Illness Research Education and Clinical Center MIRECC) at the Philadelphia VAMC (David Oslin, MD, PI) and NIDA Grant P60 DA 05186 (P.I.: Charles O’Brien) and P50 DA012756 (P.I.: H. Pettinati). We would like to acknowledge NIDA's Center for Genetic Studies in conjunction with the Washington University and the Rutgers University Cell & DNA Repository for providing DNA samples collected from the following studies and investigators: Opioid Samples: Addictions: Genotypes, Polymorphisms and Function, Mary Jeanne Kreek, MD; Genetics of Opioid Dependence, Joel Gelernter, MD, Kathleen Brady, MD, PhD, Henry Kranzler, MD, Roger Weiss, MD; Opioid Dependence, Wade Berrettini, MD, PhD. Cocaine Samples: An Introduction to the Family Study of Cocaine Dependence, Laura Bierut, MD; Genetics of Cocaine Induced Psychosis, Joseph F. Cubells, MD, PhD. The NIMH control subjects were collected by the NIMH Schizophrenia Genetics Initiative ‘Molecular Genetics of Schizophrenia II’ (MGS-2) collaboration. The investigators and coinvestigators are ENH/Northwestern University, Evanston, IL, MH059571, Pablo V. Gejman, MD (Collaboration Coordinator; PI), Alan R. Sanders, MD; Emory University School of Medicine, Atlanta, GA, MH59587, Farooq Amin, MD (PI); Louisiana State University Health Sciences Center; New Orleans, Louisiana, MH067257, Nancy Buccola APRN, BC, MSN (PI); University of California-Irvine, Irvine, CA, MH60870, William Byerley, MD (PI); Washington University, St. Louis, MO, U01, MH060879, C. Robert Cloninger, MD (PI); University of Iowa, Iowa, IA, MH59566, Raymond Crowe, MD (PI), Donald Black, MD; University of Colorado, Denver, CO, MH059565, Robert Freedman, MD (PI); University of Pennsylvania, Philadelphia, PA, MH061675, Douglas Levinson, MD (PI); University of Queensland, Queensland, Australia, MH059588, Bryan Mowry, MD (PI); Mt. Sinai School of Medicine, New York, NY, MH59586, Jeremy Silverman, PhD (PI).